Professional thesis
Open access

Can Small Banks Compete on Local Data in Online Retail Loans? The Example of Baotou Rural Commercial Bank

Other title“市民贷” 地方中小银行线上零售贷 款业务应用研究
ContributorsChen, Yunxiang
DirectorsHau, Harald
Number of pages52
Imprimatur date2020
Defense date2020

At present, more than 2000 local small- and medium-sized banks in China are facing fierce competition in terms of retail loans between online internet company and offline homogeneous banks. However, these small and medium-sized banks are not lack of opportunities, but lack of capacity. With the rapid development of Internet technology, small and medium-sized banks can gain competitive advantage by using the local big data resources which are relatively easy to obtain.

Through the application of an online ‘Citizen loan’ personal credit product of Baotou rural commercial bank (BTB), this paper deeply studies and analyzes all kinds of data fusion and modeling available to small and medium-sized banks, and establishes a localized ‘Citizen loan’ risk control model by using XGBoost and "Knowledge Graph" tools. In this practice process, this paper deeply compares and analyzes the advantages and disadvantages of the models based on public data sets and local data sets (superimposing the former), and concludes that the optimization local model based on local data sets has better performance. This ‘Citizen loan’ risk control model based on local data set was actually deployed and applied in BTB in 2019, and has greatly improved both customer benefits and bank benefits.

This result fully shows that the research results of this paper can help small and medium-sized banks to find a feasible way to realize the transformation of retail strategy through financial technology empowerment.

  • Citizen Loan
  • Credit Risk Modeling
  • Data Feature Engineering
  • XGBoost
  • Knowledge Graph
Citation (ISO format)
CHEN, Yunxiang. Can Small Banks Compete on Local Data in Online Retail Loans? The Example of Baotou Rural Commercial Bank. 2020.
Main files (2)
Thesis - Version chinoise
  • PID : unige:177595
  • Thesis number : 0028

Technical informations

Creation06/04/2024 9:19:59 AM
First validation06/05/2024 7:43:30 AM
Update time06/05/2024 7:43:30 AM
Status update06/05/2024 7:43:30 AM
Last indexation06/05/2024 7:43:38 AM
All rights reserved by Archive ouverte UNIGE and the University of GenevaunigeBlack